Skip to content
Open source platform for the machine learning lifecycle
Branch: master
Clone or download
smurching Add metric step support to OSS AbstractStore implementations & server (

Adds support for specifying metric step (x coordinate) to OSS AbstractStore implementations: FileStore, SQLAlchemyStore, RestStore, as well as the OSS server.

This commit enables users to submit LogMetric and LogBatch REST API requests containing metric x-coordinates to the OSS server. However, there is effectively no user-facing Python API that enables logging metric x coordinates (it's possible via the low-level MlflowClient.log_batch method, which accepts mlflow.entities.metric instances, but not via MlflowClient.log_metric or the fluent mlflow.log_metric, mlflow.log_metrics APIs).

Full client/fluent API support for logging x coordinates in Python will be included in a follow-up PR.
Latest commit 89bba8a Apr 20, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
.github Very minor fix to the PR template Apr 19, 2019
docs [enhancement] Hadoop artifact repository with Kerberos authorization … Apr 19, 2019
examples Expose RunData fields (metrics, params, tags) as dictionaries in Pyth… Apr 16, 2019
tests Add metric step support to OSS AbstractStore implementations & server ( Apr 19, 2019
.dockerignore Add missing files (.dockerignore, .travis.yml) (#12) Jun 7, 2018
.gitignore Debug slow pip installations breaking our Travis CI builds (#515) Sep 19, 2018
.travis.yml Fix R travis build Apr 9, 2019
CHANGELOG.rst Add changelog for Python release (#1115) Apr 10, 2019
CONTRIBUTING.rst Style edit and fix some formatting bugs. (#862) Feb 24, 2019
Dockerfile add JRE to image (#816) Jan 21, 2019 Recommend a simpler command to get version (#107) Jul 4, 2018
LICENSE.txt Update LICENSE.txt (#29) Jun 8, 2018
README.rst [Default envs] Add docker/sagemaker integration tests with default co… Nov 14, 2018
dev-requirements.txt Run MLProjects on docker containers (#555) Jan 18, 2019 Java SDK for MLflow (#380) Aug 28, 2018 Log the run page URL as a tag when running on Databricks (#388) Aug 29, 2018
pylintrc Updates to Projects API (#82) Jul 18, 2018 [enhancement] Hadoop artifact repository with Kerberos authorization … Apr 19, 2019 Include protobuf dependencies and use C implementation. (#74) Jun 27, 2018


MLflow Beta Release

Note: The current version of MLflow is a beta release. This means that APIs and data formats are subject to change!

Note 2: We do not currently support running MLflow on Windows. Despite this, we would appreciate any contributions to make MLflow work better on Windows.


Install MLflow from PyPi via pip install mlflow

MLflow requires conda to be on the PATH for the projects feature.

Nightly snapshots of MLflow master are also available here.


Official documentation for MLflow can be found at


To discuss MLflow or get help, please subscribe to our mailing list ( or join us on Slack at

To report bugs, please use GitHub issues.

Running a Sample App With the Tracking API

The programs in examples use the MLflow Tracking API. For instance, run:

python examples/quickstart/

This program will use MLflow Tracking API, which logs tracking data in ./mlruns. This can then be viewed with the Tracking UI.

Launching the Tracking UI

The MLflow Tracking UI will show runs logged in ./mlruns at http://localhost:5000. Start it with:

mlflow ui

Note: Running mlflow ui from within a clone of MLflow is not recommended - doing so will run the dev UI from source. We recommend running the UI from a different working directory, using the --file-store option to specify which log directory to run against. Alternatively, see instructions for running the dev UI in the contributor guide.

Running a Project from a URI

The mlflow run command lets you run a project packaged with a MLproject file from a local path or a Git URI:

mlflow run examples/sklearn_elasticnet_wine -P alpha=0.4

mlflow run -P alpha=0.4

See examples/sklearn_elasticnet_wine for a sample project with an MLproject file.

Saving and Serving Models

To illustrate managing models, the mlflow.sklearn package can log scikit-learn models as MLflow artifacts and then load them again for serving. There is an example training application in examples/sklearn_logisitic_regression/ that you can run as follows:

$ python examples/sklearn_logisitic_regression/
Score: 0.666
Model saved in run <run-id>

$ mlflow sklearn serve -r <run-id> -m model

$ curl -d '[{"x": 1}, {"x": -1}]' -H 'Content-Type: application/json' -X POST localhost:5000/invocations


We happily welcome contributions to MLflow. Please see our contribution guide for details.

You can’t perform that action at this time.